1975
Acknowledgments. The authors thank Dr. Morton J. Gibian for helpful discussions, and Messrs. David N. Roark and William Snider for experimental
assistance. Support from the National Institutes of Health (General Medical Sciences; G M 13342) is gratefully acknowledged.
The Crystal Structure of a Carbanion. Potassium 4,4-Dinitro-2-butenamide James R. Holden’ and Charles Dickinson Contribution from the U.S. Naoal Ordnance Laboratory, White Oak, SiIver Spring, Maryland. Received March 4, 1967 Abstract: The crystal structure of the potassium salt of the 4,4-dinitro-Zbutenamide carbanion has been determined by X-ray diffraction. The Entire carbanion is essentially planar. The N-C-N bond angle is !20° in spite of an 0-0distance of only 2.51 A between the nitro groups. The carbanions lie in planarJayen 3.0 A apart and are doubly hydrogen bonded into chains within these layers. The potassium cations lie 0.7 A above and below the anion planes.
N
itromethane is a much stronger carbon acid than methane. However, unlike other acidifying groups, the effects of a second and third nitro group are not additive (see Table I). This lack of additivity has been attributed to steric repulsions which prevent all the atoms of the resulting carbanion from lying in the same plane.2 Table I.
pK, in Water at
CH3NOa CHZ(N0z)z CH(NO,)x
See ref 2.
11 4 0
2 5 ” O
CHaCN CHz(CN)a CH(CN)x
25 12
-5b
CHSOZCHI 23 CH~SOZCHI)~ 14 CH(S02CHa)a 0
* R. H. Boyd, J. Phys. Chern., 67,737 (1963).
The thermodynamic acidity of a carbon acid (measured by the pK,) is a function of the relative stability of the undissociated molecule and of the carbanion which results from deprotonation. A nitro group increases the acidity of methane both by stabilizing the carbanion and by decreasing the stability of the neutral molecule. Its stabilizing effect on the methide ion is due chiefly to delocalization of the charge by p-orbital interaction. Maximum interaction requires that the p orbitals of the carbon and nitrogen atoms be parallel, that is, that the sp2 hybrid orbitals be coplanar. Therefore, two or three nitro groups can exert their full stabilizing effects only if the entire carbanion is coplanar. Nitro groups make methane less stable toward deprotonation by withdrawing electrons from the carbon atom and thus weakening the C-H bond. Trinitromethane (nitroform) should be less stable than carbomethoxydinitromethane (methyl dinitroacetate) because the nitro group has a stronger electron-withdrawing effect than the carbomethoxy group. This estimate is based on the Q* values of 2.00 listed by Taft3 for carbomethoxy and 3.9 which can be derived for nitro (1) Author to whom inquiries should be addressed.
(2) D. J. Cram, “Fundamentals of Carbanion Chemistry,” Academic Press Inc., New York, N. Y., 1965. (3) R. A. Taft in “Steric Effects in Organic Chemistry,” M. S. Newman, Ed., John Wiley and Sons, Inc., New York, N. Y., 1956, p 619.
from other values given by Taft.3 From other values given in this compilation,3 we estimate a Q* value of 0.6 for the carbomethoxyvinyl group ; therefore, carbomethoxyvinyldinitromethane (methyl 4,4-dinitro-2-butenoate) should be more stable than methyldinitroacetate and much more stable than nitroform. As would be predicted from the proposed stabilities of the neutral molecules, nitroform is about 1.0 pK unit stronger an acid than methyl dinitroacetate. However, 4,4-dinitro-2-butenoic acid derivatives are as strong or stronger acids than nitroform (see Table 11). This would indicate that these compounds form more stable carbanions than either nitroform or methyl dinitroacetate. Table 11. pK. in Methanol4
(NO&CHCH=CHCOOCH3 (NO&CHCH=CHCN (NOa)3CH
3.15
1.91 3.22
As stated above, a major factor in carbanion stability is delocalization of the charge from its formal location on the carbon atom. The reactivity of a carbanion toward an a,@-unsaturated system in a Michael-type addition gives an indication of the extent of such delocalization because the nucleophilicity of the carbanion is a function of the charge remaining on the carbon atom. Both trinitromethide and carbomethoxydinitromethide react readily with methyl acrylate in a Michaeltype addition, but carbomethoxyvinyldinitromethide is totally unreactive. 4 , 5 This is additional evidence for the greater stability of the carbomethoxyvinyldinitromethide carbanion and indicates that the reason for its stability is extensive charge delocalization. An obvious explanation for this carbanion stability (due to efficient charge delocalization) would be that 4,4-dinitro-2-butenoic acid derivatives form planar carbanions permitting maximum resonance interaction by (4) L.A. Kaplan, private communication. ( 5 ) L. A. Kaplan and D. J. Glover, J . A m . Chern. Soc., 88, 84 (1966).
Holden, Dickinson
Potassium 4,4-Dinitro-t-burenamide
1976
both nitro groups and by the double bond. However, this conclusion does not agree with that drawn by Kamlet and Glover6 from a study of the ultraviolet spectra of these compounds. A determination of the crystal structure of the potassium salt of 4,4-dinitro-2butenamide was undertaken to determine whether the anion does indeed assume a planar configuration. From such a structure we hoped to learn something about the balance between the steric effects (favoring nonplanarity) and the electronic effects (favoring planarity). The amide was chosen for crystallographic convenience since it forms a stable crystal with a unit cell of desirable size and shape. Experimental Section A sample of potassium 4,4-dinitro-2-butenamide containing suitable crystals was prepared as described p r e v i ~ u s l y . ~The dimensions of the triclinic cell were determined from oscillation and Weissenberg films with the crystal rotatFd about the a and F axes: a = 4.557 i 0.01 A, b = 8.311 & 0.01 A, c = 9.898 =k 0.01 A, a = 96” 3’ f 20’, p = 87” 12’ f 20’, y = 100” 55’ f 20’. These values give a calculated density of 1.93 g/cc with two molecules per cell. The density as determined by flotation in a methylene iodidetetrachloroethylene solution is 1.90 g/cc.B Intensities were collected with the crystal rotating about the a axis. Levels Ok/ through 4kl were collected using a Nonius integrating Weissenberg camera operating in a two-directional integration mode. Four-film packs were used and levels l k l through 4k/ were collected at relative spindle axis settings of 0 and 180” to eliminate the need for an oversize integration area to accommodate the expanded spots on half of the film. The intensities were read using a semiautomatic microdensitometer designed and built at this laboratory. The densitometer scanned in the 0 direction, and the optical density difference between the peak height and interpolated background was correlated with exposure by means of a known gray scale. Of the 1479 reflections in the region of reciprocal space measured (Cu Ka, 1.5418 A), 1095 had measurable intensity.
Determination of Structure All calculations for the determination of this structure were performed on an IBM 7090 using the crystallographic computing system X-ray 63.9 The intensities within each film pack were correlated giving greater weight to those measurements in the middle of the intensity range. The two film packs for each of the levels lkl through 4kl were correlated using reflections common to both packs. The intensities were corrected for absorption by Burnham’s method. lo These corrected intensities were then converted to structure factors. Quasi-normalized structure factors (Karle-Hauptman E’s)” were calculated using form factors for the neutral species C, 0, N, and H and the ionic species K+ as given in the “International Tables for X-Ray Crystallography.”’* The average values for /El and lE* -- 11 from this calculation (Table 111) indicated the centric space group P i . A sharpened, origin-removed Patterson calculation from the normalized structure factors ( € E 2- 1) clearly indicated the position of the potassium ion. An electron density map using phases determined from the po-
Table 111. Average Values of Normalized Structure Factors ~~~
Observed Theoretical for centric11 Theoretical for acentric”
E*
( E 2 -11
0.814 0.798 0.886
1 .Ooo 1.OOO 1 .Ooo
0.939 0.968 0.736
tassium position indicated positions for the nonhydrogen atoms of the carbanion. Fourier and full-matrix least-squares refinement using individual, isotropic temperature factors produced a structure with an agreement factor ( R ) of 0.15. At this point, reflection (112) was omitted from the refinement. Although this reflection had the largest observed intensity, its calculated structure factor was more than twice its observed value. The reflection intensity is either measured too small because it exceeds the response of the film or is reduced by secondary extinction. Least-squares refinement with anisotropic temperature factors then lowered R to 0.089. The four hydrogen atoms of the carbanion were placed by assuming that all of the atoms are trigonal, of the adjathat H1 lies on the line between N1 and 01’ cent carbanion (a linear hytrogen bond), and that the bond distances are all 1.075 A. Inclusion of the hydrogen atoms lowered R to 0.085, but their positions were not stable when an attempt was made to include them in the least-squares refinement. Therefore the final four cycles of refinement were carried out with the hydrogen positions generated as indicated by the positions of the heavy atoms. In the final cycle the R value was 0.085, and the maximum shift/error ratio was 0.0009, The data were grouped by Weissenberg level (all reflections with the same Miller index h) and a scale factor assigned to each level. These level scale factors were determined during the early cycles of least-squares refinement using isotropic temperature factors. They were not included in the final refinement because of the high correlation which would exist between the scale factors and the BI1component of the anisotropic temperature factors. l 3 This procedure could introduce or mask an over-all anisotropic trend in the temperature factors; therefore, any such trend may have no physical reality. In all least-squares refinements, the quantity minimized was Zw(F, - F J 2 , where w is the weight assigned to the reflection. For unobserved reflections, (Fc Fmi,)*was included in the sum when the calculated structure factor, F,, was greater than Fmi,,the structure factor calculated from the minimum observable intensity. No contribution was included for unobserved reflections when F, was smaller than Fmi,,. The weight was 1.0 for reflections with Frel less than 10.0 and (10.0/F,e1)2 for reflections with Frelgreater than 10.0, The final atomic parameters are given in Table IV. The numbers in parentheses are the errors in the last digits as indicated by the inverse matrix from the last least-squares cycle. The anisotropic temperature factors are of the form
(6) M . J. Kamlet and D. J . Glover, J . O r g . Chem., 27, 537 (1962). (7) M . J . Kamlet and L. A . Kaplan, ibid., 28, 2128 (1963). (8) H. T. Simmons, private communication. (9) J. M. Stewart, et al., Technical Report TR-64-6, NsG-398, Computer Science Center of the University of Maryland, 1964. (10) C. W. Burnham, A m . Mineralogist, 51, 159 (1966). (11) I. L. Karle and J. Karle, Acta Cryst., 16,969 (1963). (12) “International Tables for X-Ray Crystallography,” Vol. 111,
Kynoch Press, Birmingham, England, 1962.
Journal of the American Chemical Society
I El
+
+
+ +
exp[- 1/4(h2a*2Bll k2b*2B22 12c*2B3s 2hka*b*Blz 2hla*c*B13 2klb*c*Bz3)]
+
These B t j values are on the same scale as the isotropic B values listed for the hydrogen atoms. (13) E.C . Lingafelter and J. Donohue, Acta Cryst., 20, 321 (1966)
1 90:8 1 April I O , 1968
1977 Table IV.
Atomic Parameters B or Bll
B33
Biz
0.3706(3)
0.1161 (2)
0.3592(1)
2.71 (6)
2.69(6)
2.26(5)
1.13(5)
-0.72(4)
-0.13(4)
Cq
-0.0862(13) -0.1674(13) -0.0263(12) -0.0499(12)
0.2260(7) 0.3873(6) 0.4761 (6) 0.6339(6)
0.0863(6) 0.1329(5) 0.2404(6) 0.3069(6)
1.79(25) 2.14(25) 1.80(24) 1.70(23)
1.42(20) 1.24(19) 1.24(19) 1.29(19)
2.26(22) 1.84(21) 1.95(21) 1.91 (21)
0.83(17) 0.99(17) 0.62(16) 0.65(16)
-0.09(17) -0.28(17) -0,l0(16) -0.38(16)
-0.36(16) -0.40(15) -0.32(16) -0.56(16)
N1 NP N1
-0.2257(12) 0.1270(11) -0.2350(11)
0.1497(6) 0.6897(6) 0.7320(6)
-0.0236(5) 0.4184(5) 0.2632(5)
2.57(24) 2.03(21) 1.65(20)
1.40(18) 1.28(17) 1.35(17)
2.73(21) 1.74(18) 2.41 (20)
1.03(16) 0.57(14) 0.73(14)
-0.90(17) -O.l0(14) 0.19(15)
-1.00(15) -0.86(14) -0.31 (15)
01
0.0999(12) 0.2994(10) 0.1168(11) -0.2486(11) -0.3911 (10)
0.1690(5) 0.6032(5) 0.8265(5) 0.8715(5) 0.6726(5)
0.1435(4) 0.4543(4) 0.4856(5) 0.3177(5) 0.1642(5)
4.75(26) 2.64(20) 3.41 (23) 4.07(25) 2.42(20)
1.71(17) 2.12(17) 1.97(18) 1.57(17) 2.26(18)
2.61 (19) 2.14(17) 3.24(21) 3.84(22) 3.15(20)
2.00(17) 1.20(14) 1.21 (15) 2.03(16) 1.09(15)
-1.53(17) -0.55(14) -1.04(16) -1.10(18) -1.03(15)
-1.15(14) -0.29(13) -1.79(15) -1.25(15) -0.61 (15)
-0.1795 -0.3864 -0.3344 0.1325
0.0328 0.2053 0.4324 0.4174
X
K+ Ci
Cs Ct
0 2 0 3
Oq
0: Hi
Hz Ha Hq
z
Y
-0.0676 -0.0690 0.0822 0.2839
BZz
B13
B2a
2.20 2.20 2.20 2.20
A table of observed and calculated structure factors has been deposited with the American Documentation Institute. l 4
Discussion A significant feature of the crystal structure of potassium 4,4-dinitro-2-butenamide is the planarity of the anion in spite of what should be large steric effects. The equation of the least-squares planeI5 through the carbon, nitrogen, and oxygen atoms listed in Table IV is
+
- 2 . 9 0 3 ~ - 2 . 9 8 8 ~ 6.0302 = 0.0897 where x, y , and z are fractional coordinates in the triclinic cell. The staadard deviation of these atoms from the plane is 0.027 A. The distances (in angstroms) of the individual atoms of the anion from the plane are C1 through C4: -0.006, -0.040, -0.014, -0.011; N1 through N3: 0.024, 0.003, 0.008; and 01 through Os:0.019, 0.022, -0.030, 0.057, -0.026 (see Figure 1 for atom designations). Since the maximum estimated error in the position of an atom is 0.006, some of the deviations from coplanarity are surely real; however, we feel that no deviation is large enough to be chemically significant. The interatomic distances within the anion are given in Figure 1. The estimated standard deviations of the distances between atoms range from 0.007 to 0.009 A. The planar configuration of the anio; produces several short nonbonded distances: 2.512 A, between O3 and 04,2.653 A between C3 and 02, 2.723 A between Cs and 05,and 2.737 A between C2 and 05. The bond angles are given in Figure 2. Their standard deviations range from 0.5 to 0.6". Angle C3-C4-N3 is significantly larger than angle C3-C4-N2, and angle C2-C3-C4 is significantly larger than the normal C-C=C range, apparently to relieve the crowding in the surrounding parts of the anion. The unusually small bond angle C1-C2-C3 may result from the close packing of the anions to be discussed later. (14) The structure factor table has been deposited as Document No. 951 1 with the AD1 Auxiliary Publications Project, Photoduplication Service, Library of Congress, Washington, D. C. (15) V. Schomaker, J. Waser, R. E. Marsh, and G. Bergman, Acta Crjjst., 12, 600 (1959).
The bond lengths in the amide group, C1-O1 and CI-NI, are about average when compared to lengths of equivalent bonds reported for the compounds listed in Table V. 16-30 The lengths of bonds C1-C2 and CZ-c3 Table V.
Bond Distances in Amides (Angstroms)
C-N
c-0
Formamide Acetamide l7 Decanamide Tetradecanamide 19
1.30 1.38 1.312 1.26
1.255 1.28 1,282 1.23
Oxamidez0 Succinamide Glutaramide22 I-Adipamide23 Suberamide24
1.315 1.333 1.34 1.33 1.32
1.243 1,238 1.22 1.23 1.25
Benzamidez6 Nicotinamide z6 PicolinamideZ7 a-PyrazinamideB Glutaminez9 Glycyl- asparagine 30
1.31 1.34 1,322 1.312 1.28 1.39
1.24 1.22 1.234 1.244 1.27 1.22
are not significantly different from the values given by Dewar and Schmeising31 for single and doukle bonds between sp2 carbon atoms, 1.479 and 1.353 A, respectively. Therefore, the configuration found for the amide end of the anion is not different from that found in neutral molecules.32 (16) (17) (18) (1958). (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32)
J. Lade11 and B. Post, Acta Cryst., 7, 559 (1954). F. Senti and D. Harker, J . A m . Chem. SOC.,62, 2008 (1940). J. R. Brathovde and E. C. Lingafelter, Acta C r j > ~ t .11, , 729
J. D. Turner and E. C. Lingafelter, ibid., 8 , 551 (1955). E. M. Ayerst and J. R. C. Duke, ibid., 7 , 558 (1954). D. R. Davies and R. A. Pasternak, ibid., 9 , 334 (1956). M. Hospital and J. Housty, ibid., 21, 413 (1966). M. Hospital and J . Housty, ibid., 20, 626 (1966). M. Hospital and J. Housty, ibid., 20, 368 (1966). B. R. Penfold and J. C. B. White, ibid., 12, 130 (1959). W. B. Wright and G. S.D. King, ibid., 7, 283 (1954). T. Takano, Y . Sasada, and M . Kakudo, ibid., 21, 541 (1966). Y . Takaki, Y . Sasada, and T. Watanabe, ibid., 13, 693 (1960). W. Cochran and B. R. Penfold, ibid., 5 , 644 (1952). R. A. Pasternak, L. Katz, and R. B. Corey, ibid., 7 , 225 (1954). M. J. S. Dewar and H. N. Schmeising, Tetrahedron, 5,166 (1959). M. R. Truder, J . Chem. SOC.,997 (1960).
Holden, Dickinson j Potassium 4,4-Dinitro-Z-butenamide
I978
0
% @--a$+&
0
W
B
4.
1.233
Figure 1. Bond lengths.
c,
r
P
Bond angles.
TheSoordination of atom C4 is trigonal. It lies only 0.008 A from the plane defined by C3, N2,and N3. This is less than twice the standard deviation of the atom positions and corresponds to an angle of only 0.4" between the bonds to C4 and this plane. Therefore, the configuration at C4is not significantly nonplanar. Certainly there is no evidence in dinitrobutenamide of the 3 " nonplanarity reported for the carbanions ammonium tricyanomethide and pyridinium dicyanomethide. The two nitro groups lie close to the plane of the trigonal C4 atom. The distances of oxygen atoms O2 through 0,from the plane defined by N2, and N3 are 0.035, -0.035, 0.039, and -0.033 A. Thus the N2 and N Bnitro group planes make angles of only 1.9 and
c3,
(33) R.Desiderato and R. Sass, Acra Cryst., 18, 1 (1965). (34) C. Bugg and R. Sass, ibid., 18, 591 (1965).
Journal of the American Chemical Society
/
90:8
Arrangement of atoms within a layer ofthe structure.
2.0" with the trigonal carbon plane. These small angles indicate that the carbon and nitrogen atoms are oriented properly for p-orbital interaction, and that the nitro groups can contribute to delocalization of the charge formally assigned to C4. The fact that the bond between C2 and C3 lies close to the C4 plane indicates that CBis also oriented properly for p-orbital interaction. The length of C3-C4 is significantly shorter than the value given by Dewar and Schmeisin 3 1 for a single bond between sp2 carbon atoms, 1.479 . The lengths of C4-N2 and C4-N3 are not significantly different from each other or from the length reported for the similarly situated C-N bond in dipotassium nitroacetate. 3 5 These bonds are significantly shorter than the C-N bonds found between nitro groups and aromatic rings, which average about 1.47 A.36-39 However, they are longer than many of the C-N bonds found between amine groups and aromatic rings and those found in amide groups (Table V). We believe that the C4-N2 and C4-N3 bond lengths should be compared with those found for other nitro groups. On this basis the bond lengths given in Figure 1 indicate that the charge on the carbanion is delocalized from its formal location on C4 into the nitro groups and into the bond to CB. The crystal structure of potassium 4,4-dinitro-2butenamide shows that in the solid state the dinitrobutenamide carbanion can assume a planar configuration which allows efficient charge delocalization. If the anion also assumes this stable configuration in solution, it would explain why dinitrobutenamide is a relatively strong carbon acid. However, it would contradict some of the explanations which have been given for the nonadditivity of the effects of nitro groups based on nonplanarity of carbanions. 1 ~ 6 , 4 0 The dinitrobutenamide anions y e packed in the crystal lattice in planar layers 3.0 A apart. Figure 3 shows the arrangement of the anions within each layer. As indicated, the anions are doubly hydrogen bonded into continuous chains within the layers. Both hydro-
1
d Figure 2.
Figure 3.
(35) D. S . Sutor, F. J. Llewellyn, and H. S . Maslen, ibid., 7, 145 (1954). (36) I
